[permaculture] the next generation of Pc Design resources.

christophe mckeon gonzalez de leon chromatophore at gmail.com
Tue Nov 24 03:49:25 EST 2009


> In the case of PFAF (what an incredible resource - you should subscribe
> to their mailing list) would a user seeking information from that
> resource have access to the same search options with an OWL - PC DB that
> they would through direct access to PFAF? As I understand it PFAF offers
> a RBD user interface/front end. How would this be accessed through an
> OWL interface?

yeah the pfaf web front end is just plain html. but you might be
talking about the windows executable front end they distribute? if so
that is less than useless. they offer the db itself minus the front
end as a download which is in microsoft access format and very crufty
internally (imho) for a fee. i actually already wrote some ruby
scripts to convert the entire thing to sqlite which is much more
useful.

as far as transforming it to rdf triples, there would be various ways
of approaching this, the best being to get permission to copy the
entire db over to some triplestore and to serve it up that way on the
web through a sparql endpoint and as html for users to update etc.
that would be what we would ideally want to do if the database were to
become open for many people to update. right now it's totally silo-ed.

actually i just reread their license, and as long as the site we are
running is not commercial and we share our additions and attribute the
authors, then we can do what we like with the data. they consider an
advertising driven site commercial. if the site were owned by a
non-for-profit and had advertising as a means of funding i think that
the pfaf people would probably be cool with that but we'd have to ask.

> - this seems like an opportune time to utilize the added capability of
> OWL to go beyond the PFAF user interface and database engine to relate
> and integrate PFAF data with that in other DB datasets on the Web or
> other data sources in text format or

yep that's the idea. especially as data available in rdf becomes more
and more prevalent adding disparate datasources to the project will be
much more easily done if we are using semweb technologies. also if we
publish through a sparql endpoint then everything we have compiled
becomes grist for somebody else to very easily mashup.

>... Think of an online knowledge system that becomes a global tool for
> anyone to use for site development and maintenance, one that allows them
> to enter, store, process and retrieve data about their site and region,
> i.e. their homestead or farm, their community or town and their
> bioregion.

that's pretty much exactly what i have in mind as well. the main thing
is that it must really be a useful and easy to use tool or else there
is little incentive for people to enter their own data. i'm thinking
two core predicates in a permaculture ontology can go a very long way,
namely 'needs' and 'yields'. if most resources in a db can have rdf
statements of the form x needs y and y yields z, then some pretty
powerful and not so complicated reasoners can be perched on top of
that to make suggestions and connections.

right now for example you have to go through the pfaf database and
make all the connections yourself. what if you looked up a plant, and
the system automatically showed you how that plant could fit in with
what you already have going on your site.

as an example, as a user you might have already made these rdf
statements (through a much less formal interface of course)

rdf statements or triples are in this form: subject predicate object

i haveTooMany chickens
i haveA pond
i need firewood
i liveIn "wonderland"

the system armed with some other triples somebody else has already encoded:

haveTooMany isEquivalentTo yields
chickens yield eggs
eggs yield eggShells
eggShells yield calcium
crayfish need calcium (for their shells after their molt)
james livesIn "wonderland"
james needs chickens
james hasTooMany trees
trees yield firewood
crayfish need ponds

the system could then infer a great deal as is plain to see. some
suggestions it could make are

that james who lives right near by might have firewood in exchange for
chickens.
that crayfish might be worth considering (because chickens yield eggs
which yield calcium which are good for crayfish and crayfish need
ponds which you already have).
etc..

this is with a tiny dataset and a poorly thought out ontology. what
about a huge dataset of triples and
a well thought out and much richer ontology. such a system could
potentially surprise even veteran
permaculturists with novel connections.

_c



More information about the permaculture mailing list